| From: | Stephen Frost <sfrost(at)snowman(dot)net> |
|---|---|
| To: | Stephan Szabo <sszabo(at)megazone(dot)bigpanda(dot)com> |
| Cc: | Per Jensen <per(at)net-es(dot)dk>, Postgres general list <pgsql-general(at)postgresql(dot)org> |
| Subject: | Re: Index scan vs. Seq scan on timestamps |
| Date: | 2004-12-07 02:49:25 |
| Message-ID: | 20041207024925.GN10437@ns.snowman.net |
| Views: | Raw Message | Whole Thread | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-general |
* Stephan Szabo (sszabo(at)megazone(dot)bigpanda(dot)com) wrote:
> On Mon, 6 Dec 2004, Per Jensen wrote:
> > select count(*)
> > from accesslog
> > where time between (timeofday()::timestamp - INTERVAL '30 d') and
> > timeofday()::timestamp;
>
> Besides the type issue, timeofday() is volatile and thus is not allowed to
> be turned into a constant in order to do an index scan because it's
> allowed to return different values for every row of the input.
Is there a way to say "just take the value of this function at the start
of the transaction and then have it be constant" in a query?
Stephen
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Andrew - Supernews | 2004-12-07 03:13:04 | Re: Index scan vs. Seq scan on timestamps |
| Previous Message | Stephan Szabo | 2004-12-07 02:34:25 | Re: Index scan vs. Seq scan on timestamps |